Return to the Little Kingdom

by Michael Moritz

  Named after the short, narrow skirts made popular by London’s Carnaby Street, the minicomputers were usually no larger than a combination refrigerator-freezer for a family of six. The minicomputer makers, just like the companies that designed satellites and rockets, capitalized on the great shrinking world of electronics. As the semiconductor companies developed their manufacturing techniques, they squeezed more and more transistors onto single pieces of silicon. This made it possible for companies like Digital Equipment to produce computers that, even if they didn’t match the performance of a contemporary mainframe, were more powerful than some of the mainframes that had been made five years earlier. Every graph that appeared in the trade magazines and plotted price against performance showed the machines would become still cheaper and even more powerful.

  But even though minicomputers were far smaller than mainframes, they still needed bulky attachments. Programs were entered on paper tape; the memory was formed out of dozens of small doughnut-shaped pieces of iron linked by wires and built into blocks that were the size of cigar boxes. Results of programs appeared on a Teletype printer. The handbooks and manuals revealed something of the complexity of trying to control the flow of millions of bits moving in all sorts of directions. Digital Electronics Corporation’s Small Computer Handbook,which the Sylvania analysts gave to Wozniak, became something of an industry classic because it revealed so much about the computer. It included detailed descriptions of the quirks of the central processing unit, provided directions about how the memory should be managed, presented ways of making connections with the Teletype machine, and provided flow charts to help with the writing and testing of programs.

  The computer trade magazines were accompanied by a more specialized literature: the component magazines. At the end of the sixties these focused on the integrated circuits, the chips made by semiconductor companies like Fairchild, Signetics, Synertek, Intel, and Motorola. For Wozniak and Baum these magazines became almost as important as the computer magazines and computer manuals. Though no semiconductor company was making a single chip that performed like a computer, some did make chips that, with sufficient ingenuity, could be combined to act as a computer. The companies themselves released details of the features and performance of their new chips on what were called data sheets which were chockablock with technical information. These too became sought-after items. Designing a decent computer, a computer that approached that distant world, the state of the art, required intimate familiarity with the diagrams and details of the data sheets.

  Though he pored over DEC’s Small Computer Handbook, the Varian 620i was the first minicomputer Wozniak subjected to close inspection. It was packed into a brown cabinet with rows of black and white switches on the front panel. For the first time, Wozniak tried to design his own version of a minicomputer with chips he selected: “I didn’t know how to make a complete computer but I understood what a computer was.” He began to understand the layers between the program that a user would type into a computer and the very heart of the machine. He focused on the heart and understood the idea of a set of precise instructions that formed a code to control the machine.

  But if he had not mastered all the links of computer design, Wozniak had fastened onto the idea of using as few parts as possible. He was delighted when he discovered a way of combining or eliminating gates, the circuits that form the basis of digital logic. When chips contained circuits that would replace several gates they became the cause of jubilation. Wozniak began to concentrate on making parts perform as many functions as possible. “I started moving toward higher levels of integration.” Both Wozniak’s and Baum’s parents were startled by their sons’ progress. Like many other teenagers they were free from life’s dreary distractions and had the luxury of enough time to pursue their obsessions.

  Wozniak and Baum soon sorted out their favorite minicomputers and their bedroom bookshelves began to bulge with computer pamphlets. They started to differentiate among computers, between clever and clumsy designs. They appreciated abstruse features like the way in which some machines handled floating decimal points. Occasionally a name, or cosmetic appeal, tickled their fancy like the Skinny Mini which was named for its thin cabinet. Elmer Baum said, “After about three months I gave up. They were designing computers and I couldn’t understand what they were talking about.”

  When Wozniak left high school for college he took his interests with him. He was rejected by his father’s alma mater, Cal Tech, and after a miserable day at De Anza Community College in Cupertino, he enrolled at the University of Colorado in Boulder. Jerry Wozniak viewed his son’s attempts to abandon California and join some high-school friends with suspicion. “Stephen wasn’t ready to leave home and go off to college at the same time.” One of the items packed into his suitcase was an oscillator that had been specially tuned to jam television reception. Wozniak started to interfere with closed-circuit lectures, provoking the professors to try to adjust the television set. He kept twiddling his oscillator until the teachers were in contortions, convinced that if they kept an arm or leg in the air the interference would disappear. He also managed to infuriate some classmates by jamming a transmission of the Kentucky Derby just as the horses were about to cross the line.

  Wozniak’s life at Colorado revolved around the University’s Control Data computer, CDC 6400. He read the computer’s manuals, learned some more techniques in FORTRAN programming and also became familiar with another computer language, ALGOL. For the college administrators Wozniak was a nuisance who spent too much time hanging around the computer room and far too much time using the computer. He ran a couple of programs that spat out reams of paper saying: FUCK NIXON and GOOD SCRAP PAPER. “I was spending ten hours on computers for every hour of class.” Late-night bridge sessions and hundred-mile jaunts for hamburgers didn’t help his academic performance either. He was badgered by one of the deans and threatened with expulsion. Wozniak retaliated by hiring a lawyer to write a threatening letter but that hardly improved matters. At the end of his first year he left Colorado with a suitcase full of more-refined computer designs and a bundle of Fs, and returned to his parents’ home in Sunnyvale where he enrolled once more at De Anza Community College.

  Home again, Wozniak was pulled back into the same small circle and the milieu of cosmetic rejects, data sheets and science fairs. He and Allen Baum took some of the same classes at De Anza while Elmer Baum also enrolled in a course that taught the programming language FORTRAN. After a few weeks he dropped out and his admiration for the skills of the younger pair increased. Wozniak ran afoul of more teachers as he toyed with computer designs during linear algebra lessons.

  At the end of the year he and Baum accidentally found summer work. They were out looking for the local office of a minicomputer company when they strolled into the headquarters of Tenet, a small company that was trying to make computers for customers like the California Department of Motor Vehicles. The pair talked themselves into jobs as programmers and though Baum soon left to start his studies at MIT, Wozniak stuck it out and learned how to program a computer system that could serve many users simultaneously. He made the occasional jaunt to Los Angeles—“I wanted to marry my young cousin down there but she never liked me”—stayed at Tenet until it fell victim to the 1972 recession, and then registered for unemployment benefits.

  Meanwhile, he was learning, in a haphazard manner, much more about computer design. He read the Xerox copies of computer textbooks that Baum mailed from MIT and he still visited school science fairs. During one visit he spotted an enlightening entry. The item that grabbed his attention was a mechanical machine that stepped, in sequence, through several instructions. At each step it was wired to fire off particular signals. Wozniak made a copy of the writeup that accompanied the machine and took it home to read. He translated the concept into electronics and grasped the idea of a circuit that would step through many small operations in an ordained sequence before performing an instruction: “All of a sudden I understood
sequencing steps. I knew immediately that I knew how to design computers and I hadn’t the day before. You just know it. As soon as a good concept clicks you just know that it got you there.”

  The self-taught lesson was of considerable help when Wozniak delved into the innards of Data General’s Nova minicomputer. Designed by a team of refugees from Digital Equipment Corporation, the Nova gained a reputation for clever and aggressive design. A fancy poster that the company mailed out was a sought-after item in the small world of camp followers. Wozniak and Baum both hung the poster among the parade of idols that decorated their bedroom walls and the former explained the attraction. “There was no other computer around that looked as if it could sit on a desk.”

  The Data General Supernova was a sixteen-bit machine—it handled sixteen binary digits at a time—and everything apart from the memory was mounted on a single laminated board. Over one hundred semiconductor chips were slotted into holes in the green board and linked by squiggly solder traces. The lines of solder were etched on what was called a printed circuit board that formed one of the basic building blocks of computers. The chips mounted on the “mother board” controlled the most important functions of the machine. Almost every aspect of the Data General computer provided some commentary on the progress of electronics. Though the computer’s arithmetic logic was far more sophisticated, it was still akin to the adder-subtracter Wozniak had designed when he was thirteen. But what had, in 1963, required a large board and hundreds of parts was contained on a sliver of silicon in 1970.

  Along with Baum, who spent his summer vacations in California, Wozniak started to design his own version of the Nova. He wrote to Data General asking for more information and received several hundred pages of internal company documents. The pair gathered data sheets on new chips made by Fairchild Semiconductor and Signetics, pored over the technical specifications, and selected the chips that suited their needs. They drew schematics—diagrams that illustrated how the chips would be linked—for a couple of different versions of the computer. One used chips made by Fairchild; the other used chips made by Signetics.

  Though Wozniak was the driving force, Baum was more than a cheerleader. He was familiar with every aspect of the design and would suggest how the maximum amount of power could be extracted from the chips. They concentrated on the digital electronics and shrugged off more humdrum concerns. Baum recalled: “We didn’t worry about things like the power supply.” At one stage the pair even considered building their own version of the computer, filled a folder with schematics, and wrote to companies asking for parts. Wozniak recalled: “Every computer I designed I intended to build. Getting the parts was the problem.”

  The rigor of designing several different versions of the Nova provided Wozniak with some illuminating lessons. To help his son understand some subleties, Jerry Wozniak arranged for him to meet the designer of a Fairchild semiconductor chip. The Fairchild engineer explained that the number of chips used in a design formed only one aspect of the final goal. He told Wozniak that the space occupied by the chips on a printed circuit board mattered just as much as the number of chips. Henceforth Wozniak focused on the twin objectives of combining as few chips as possible in the smallest amount of space.

  The experience with the Data General Nova prodded Wozniak toward a grander diversion. He decided to try to build his own computer. He managed to spur the interest of one of his neighborhood chums, Bill Fernandez, to help with the effort. They had known each other for several years and their fathers played golf together. Though several years younger than Wozniak, Fernandez with his tense, thin, ivory features had a broader range of interests. He became a member of the Bahai faith, studied aikido, and seemed like the sort of person who might have been at home in sixteenth-century Japan as the student of a samurai warrior. He too was lured by science fairs and one year had entered an electric lock that had flat switches nailed to a piece of plywood. He built sirens from oscillators and was, as he readily admitted, thorough and competent but not given to whim or impulse. He was fastidious, good with his hands, and had a knack for installing items like car radios.

  In his final year at McCollum’s electronics class, Fernandez had worked as a technician in NASA’s spacecraft-systems laboratory. There he built, tested, and modified circuits, learned about special soldering techniques, was taught how to dress leads properly, and was lectured on the perils of nicking wires. Fernandez carved out a corner of his parents’ garage to work on his hobby. He squeezed his own shelves and workbench between the family water heater and clothes dryer. “Space in the garage was a constant battle. They were saying I had a quarter of the garage when I only had a sixteenth.” But the Fernandez garage offered a sturdy place to build Wozniak’s machine.

  Wozniak knew what he wanted from his computer. “I wanted to design a machine that did something. On a TV you turn a knob and it does something. On a computer you push a button and some lights come on.” To build a machine that would blink, Wozniak and Fernandez started scavenging for parts from a bundle of semiconductor companies. Intel furnished them with eight memory chips each of which could store 256 bits. Intersil gave a couple of expensive chips that contained arithmetic logic units. They rounded up some switches from a batch of samples belonging to a salesman for a switch company, light-emitting diodes from a Monsanto engineer, and a metal frame from one of Hewlett-Packard’s scrap piles. The largest batch of parts came from a couple of applications engineers at Signetics. Wozniak and Fernandez spread their trophies out on the latter’s living-room floor and sorted out all the adders, multiplexers, and registers. They checked the part numbers against the data sheets and stuffed them into rows of small, carefully labeled manila envelopes.

  Once they set to work there was a division of labor. Wozniak designed the computer on a couple of sheets of notepad paper and concentrated on the logic design. Fernandez designed the timing circuits and the circuits that hooked the computer to the lights. Wozniak watched his younger pal, who was still at high school, play technician and assemble the computer. “He didn’t really have any engineering background but he knew how to build it with straight wires and a soldering iron. He was slow but very careful and very neat.” For several weeks the pair used their evenings and weekends to build the computer and managed to swig down a considerable number of bottles of Cragmont cream soda in the process. Fernandez bicycled down to the local Safeway with the empty quarts and used the deposits to help buy the few parts they still needed.

  The Cream Soda Computer was a small version of the minicomputers that had caught Wozniak’s fancy—“It was the absolute minimum hardware”—and the design was dictated by the off-the-shelf parts in the manila envelopes. The center of the machine was formed by two four-bit arithmetic-logic units which Wozniak rigged in tandem to give an eight-bit-wide computer. The completed computer was mounted on a metal frame. One board carried the chips and a smaller one contained the timing circuits—a crystal oscillator and frequency-divider circuit adapted from a Signetics manual. Fernandez fastened eight switches into holes he had drilled in a piece of Bakelite.

  After the Cream Soda Computer was completed, Wozniak started to exercise his control over it by writing a few programs. The programs were based on the semiconductor data sheets that revealed which instructions were needed to make the chips perform functions like addition and subtraction. He listed the bits, figured out the operating code, and wrote it down. All the instructions were executed within five steps and followed an intimate sequence that Wozniak muttered to himself: “Load; load the next byte of the instruction into the memory address register; put that through the alu into the alu output register; dump the alu output register into the next memory location.”

  The timing circuit, which Fernandez designed, ensured that five signals were generated in the right order of every instruction. The programs performed actions like multiplying the values entered into four switches by the values entered into the other four switches and displayed the answer in the lights. Wozniak ref
lected on the importance of the results. “I cannot explain why that means so much to me. Multiplying two four-bit numbers by each other doesn’t sound like a lot. But being able to do something which you couldn’t have done without a computer is worth something.”

  When the computer was almost complete, Fernandez invited his friend Steven Jobs to drop by the garage, take a look at the computer, and meet its designer. Jobs was, in his own way, impressed by the machine and by Wozniak: “He was the first person I met who knew more electronics than I did.”

  Wozniak decided to reveal his computer to the world and called a San Jose Mercury reporter that his mother knew. The reporter, accompanied by a photographer, appeared in Wozniak’s bedroom for a demonstration. As Wozniak explained some of the subtleties of the ugly contraption that lay on the floor, smoke started to emerge from the power supply Fernandez had built. The computer expired as a stream of high voltage from the power supply blew out every integrated circuit on the board. Fernandez examined the power supply and found the fault lay in an unmarked chip he had earned by doing a gardening chore for one of the neighbors. He was miffed: “We didn’t get our pictures in the paper and we didn’t become boy heroes.”

  “He sells goldfish,” Goldman said.

  Inside a nineteenth-century red-brick building set among the antique stores, restaurants, and law offices of San Francisco’s Barbary Coast, a group of four men gathered to plan Apple’s advertising for the coming year. It was an afternoon of Indian summer and the conference room, which didn’t have any windows, was stuffy. The room belonged to Chiat-Day, a medium-sized advertising agency that was perpetually rumored to be on the verge of losing its largest client, Apple Computer. A couple of long-leaved potted plants sagged from the heat, a movie projector was concealed behind a sheet of smoked glass, and alongside a wet bar a refrigerator heaved like an iron lung. The four men were seated in plush chairs around a laminated conference table.